[Rd] MASS (and class, nnet, spatial) for R 2.10.0 ?
edd at debian.org
Thu Oct 15 17:22:56 CEST 2009
On 15 October 2009 at 09:54, Uwe Ligges wrote:
| The PACKAGES file in src/contrib also points to the packages in
| src/contrib/2.10.0. It is up to Dirk to use the information in PACKAGES
| correctly as it is done for the Windows and Mac binaries on CRAN.
I am -- which is why I took care of 'effects' where yesterday's release to
upgrade as MASS complained loud and clear thanks to its R (<= 2.9.2) and I am
now using R 2.10.0 beta.
The issue at hand are the 'conditional' source directories. We (eg Debian)
use so-called watch files to match current versions against http or
ftp-reachable sources, and that scheme simply does cooperate with the scheme
I could satirically describe as 'CRAN sources here when the moon is half full
but there otherwise'. *Unconditionally* computable URLs are simply easier to
handle. "We" (eg Debian) deal with some 15,000 source packages from a
presumably similarly large number of sources. Special-casing the PACKAGES
file is out of scope.
Then again, with my Debian developer hat on, I can certainly cope with that,
and I don't see a much cleaner way forward. Part of the problem simply is
that I am 'early' with some packages, but that is in simply to help in
getting pre-releases out so that we catch a bug or two. I was just trying to
point out potential sources of confusion, and you are quite right in pointing
me to the PACKAGES file.
OTOH, consider that eg Brian always reminds people to use pre-release, and if
one then does simple non-R-based searches such as the URL expansion
or the direct
one arrives at entirely misleading information. Just sayin' ... and as
above, this cannot be corrected as the Apache macro / cgi expander cannot
know that I have R 2.10.0 in mind.
I'll drop this now, and thank everybody for taking the time to follow-up.
Three out of two people have difficulties with fractions.
More information about the R-devel